Bergman Kernel and Pluripotential Theory
Zbigniew B locki
Uniwersytet Jagiello´ nski, Krak´ ow, Poland http://gamma.im.uj.edu.pl/ eblocki
Geometric invariance and nonlinear PDE
Kioloa, February 9-14, 2014
Bergman Completeness
Ω bounded domain in C
nH
2(Ω) = O(Ω) ∩ L
2(Ω) K
Ω(·, ·) Bergman kernel
f (w ) = Z
Ω
f K
Ω(·, w ) d λ, w ∈ Ω, f ∈ H
2(Ω)
K
Ω(w ) = K
Ω(w , w )
= sup{|f (w )|
2: f ∈ H
2(Ω), ||f || ≤ 1}
Ω is called Bergman complete if it is complete w.r.t. the Bergman
metric B
Ω= i ∂ ¯ ∂ log K
ΩKobayashi Criterion (1959) If
w →∂Ω
lim
|f (w )|
2K
Ω(w ) = 0, f ∈ H
2(Ω), then Ω is Bergman complete.
The opposite is not true even for n = 1 (Zwonek, 2001).
Kobayashi Criterion easily follows using the embedding ι : Ω 3 w 7−→ [K
Ω(·, w )] ∈ P(H
2(Ω)) and the fact that ι
∗ω
FS= B
Ω.
Since ι is distance decreasing,
dist
ΩB(z, w ) ≥ arccos |K
Ω(z, w )|
pK
Ω(z)K
Ω(w ) .
Some Pluripotential Theory
Ω is called hyperconvex if it admits a negative plurisubharmonic (psh) exhaustion function (u ∈ PSH
−(Ω) s.th. u = 0 on ∂Ω).
Demailly (1985) If Ω is pseudoconvex with Lipschitz boundary then it is hyperconvex.
Pluricomplex Green function For a pole w ∈ Ω we set
G
Ω(·, w ) = G
w= sup{v ∈ PSH
−(Ω) : v ≤ log | · −w | + C } Lempert (1981) Ω convex ⇒ G
Ωsymmetric
Demailly (1985) Ω hyperconvex ⇒ e
GΩ∈ C ( ¯ Ω × Ω) Open Problem e
GΩ∈ C ( ¯ Ω × ¯ Ω \ ∆
∂Ω)
Equivalently: G (·, w
k) → 0 loc. uniformly as w
k→ ∂Ω?
True if ∂Ω ∈ C
2(Herbort, 2000)
Demailly (1985) If Ω is hyperconvex then G
w= G
Ω(·, w ) is the unique solution to
u ∈ PSH(Ω) ∩ C ( ¯ Ω \ {w }) (dd
cu)
n= (2π)
nδ
wu = 0 on ∂Ω u ≤ log | · −w | + C
B. (1995) If Ω is hyperconvex then ∃! u = u
Ωs.th.
u ∈ PSH(Ω) ∩ C ( ¯ Ω) (dd
cu)
n= 1 d λ u = 0 on ∂Ω.
Open Problem u ∈ C
∞(Ω)
Pogorelov (1971) True for the analogous solution of the real
Monge-Amp` ere equation (for any bounded convex domain in R
nwithout any regularity assumptions).
B.-Y. Chen, Pflug - B. (1998) / Herbort (1999) Hyperconvex domains are Bergman complete Herbort If Ω is pseudoconvex then
|f (w )|
2K
Ω(w ) ≤ c
nZ
{Gw<−1}
|f |
2d λ, w ∈ Ω, f ∈ H
2(Ω).
Corollary lim
w →∂Ω
λ({G
w< −1}) = 0 ⇒ Ω is Bergman complete Proposition If Ω is hyperconvex then
w →∂Ω
lim ||G
w||
Ln(Ω)= 0.
Sketch of proof ||G
w||
nn= R
Ω
|G
w|
n(dd
cu
Ω)
n≤ n!||u
Ω||
n−1∞Z
Ω
|u
Ω|(dd
cG
w)
n≤ C (n, λ(Ω)) |u
Ω(w )|
Lower Bound for the Bergman Distance
Diederich-Ohsawa (1994), B. (2005) If Ω is pseudoconvex with C
2boundary then
dist
ΩB(·, w ) ≥ log δ
Ω−1C log log δ
Ω−1, where δ
Ω(z) = dist
Ω(z, ∂Ω).
Pluripotential theory is the main tool in proving this estimate, in particular we have the following:
B. (2005) If Ω is pseudoconvex and z, w ∈ Ω are such that {G
z< −1} ∩ {G
w< −1} = ∅
then
dist
ΩB(z, w ) ≥ c
n> 0.
Open Problem dist
ΩB(·, w ) ≥
C1log δ
−1ΩFrom Herbort’s estimate
|f (w )|
2K
Ω(w ) ≤ c
nZ
{Gw<−1}
|f |
2d λ, w ∈ Ω, f ∈ H
2(Ω),
for f ≡ 1 we get
K
Ω(w ) ≥ 1
c
nλ({G
w< −1}) .
To find the optimal constant c
nhere turns out to have very interesting consequences!
Herbort (1999) c
n= 1 + 4e
4n+3+R2, where Ω ⊂ B(z
0, R) (Main tool: H¨ ormander’s estimate for ¯ ∂) B. (2005) c
n= (1 + 4/Ei (n))
2, where Ei (t) =
Z
∞ tds se
s(Main tool: Donnelly-Fefferman’s estimate for ¯ ∂)
Suita Conjecture
D bounded domain in C c
D(z) := exp lim
ζ→z
(G
D(ζ, z) − log |ζ − z|)
(logarithmic capacity of C \ D w.r.t. z) c
D|dz| is an invariant metric (Suita metric)
Curv
cD|dz|= − (log c
D)
z ¯zc
D2Suita Conjecture (1972): Curv
cD|dz|≤ −1
• “=” if D is simply connected
• “<” if D is an annulus (Suita)
• Enough to prove for D with smooth boundary
• “=” on ∂D if D has smooth boundary
-5 -4 -3 -2 -1
-7 -6 -5 -4 -3 -2 -1
Curv
cD|dz|for D = {e
−5< |z| < 1} as a function of log |z|
-5 -4 -3 -2 -1
-6 -5 -4 -3 -2 -1
Curv
(log KD)z ¯z|dz|2for D = {e
−5< |z| < 1} as a function of log |z|
∂
2∂z∂¯ z (log c
D) = πK
D(Suita) Therefore the Suita conjecture is equivalent to
c
D2≤ πK
D.
Ohsawa (1995) observed that it is really an extension problem: for z ∈ D find holomorphic f in D such that f (z) = 1 and
Z
D
|f |
2d λ ≤ π (c
D(z))
2.
Using the methods of the Ohsawa-Takegoshi extension theorem he showed the estimate
c
D2≤ C πK
Dwith C = 750.
C = 2 (B., 2007)
C = 1.95388 . . . (Guan-Zhou-Zhu, 2011)
Ohsawa-Takegoshi extension theorem (1987) with optimal constant (B., 2013)
0 ∈ D ⊂ C, Ω ⊂ C
n−1× D, Ω pseudoconvex, ϕ ∈ PSH(Ω)
f holomorphic in Ω
0:= Ω ∩ {z
n= 0}
Then there exists a holomorphic extension F of f to Ω such that Z
Ω
|F |
2e
−ϕd λ ≤ π c
D(0)
2Z
Ω0
|f |
2e
−ϕd λ
0.
For n = 1 and ϕ ≡ 0 we get the Suita conjecture.
Main tool: H¨ ormander’s estimate for ¯ ∂
B.-Y. Chen (2011) proved that the Ohsawa-Takegoshi theorem
(without optimal constant) follows form H¨ ormander’s estimate.
Tensor Power Trick
We have
K
Ω(w ) ≥ 1
c
nλ({G
w< −1}) where c
n= (1 + 4/Ei (n))
2.
Take m 0 and set e Ω := Ω
m⊂ C
nm, w := (w , . . . , w ). Then e K
Ωe( w ) = (K e
Ω(w ))
m, λ
2nm({G
we< −1}) = (λ
2n({G
w< −1})
m. Therefore
K
Ω(w ) ≥ 1
c
nm1/mλ({G
w< −1}) but
m→∞
lim c
nm1/m= e
2n.
Repeating this argument for any sublevel set we get
Theorem 1 Assume Ω is pseudoconvex in C
n. Then for a ≥ 0 and w ∈ Ω
K
Ω(w ) ≥ 1
e
2naλ({G
w< −a}) .
Lempert recently noticed that this estimate can also be proved using Berndtsson’s result on positivity of direct image bundles.
What happens when a → ∞?
For n = 1 we get K
Ω≥ c
Ω2/π (another proof of Suita conjecture).
For n ≥ 1 and Ω convex using Lempert’s theory one can obtain:
Theorem 2 If Ω is a convex domain in C
nthen for w ∈ Ω K
Ω(w ) ≥ 1
λ(I
Ω(w )) ,
I
Ω(w ) = {ϕ
0(0) : ϕ ∈ O(∆, Ω), ϕ(0) = w } (Kobayashi indicatrix).
Mahler Conjecture
K - convex symmetric body in R
nK
0:= {y ∈ R
n: x · y ≤ 1 for every x ∈ K } Mahler volume := λ(K )λ(K
0)
Mahler volume is an invariant of the Banach space defined by K : it is independent of linear transformations and of the choice of inner product.
Santal´ o Inequality (1949) Mahler volume is maximized by balls Mahler Conjecture (1938) Mahler volume is minimized by cubes Hansen-Lima bodies: starting from an interval they are produced by taking products of lower dimensional HL bodies and their duals.
n = 2: square
n = 3: cube & octahedron
n = 4: . . .
Bourgain-Milman (1987) There exists c > 0 such that λ(K )λ(K
0) ≥ c
n4
nn! . Mahler Conjecture: c = 1
G. Kuperberg (2006) c = π/4 Nazarov (2012)
I
equivalent SCV formulation of the Mahler Conjecture via the Fourier transform and the Paley-Wiener theorem
I